The IV is starting and we are quite excited. A lot of us here at the MRT are involved in the organisation and we sure hope everything will go over nicely.
One of the things we look forward the most on the IV is the Demonstration Day. Demonstrations are always fun: You work towards presenting a specific state of your work, there is no hiding behind words or paper and you show off what you are really capable of. You are always at risk that something unexpected happens: broken parts, external conditions that you didn't account for, bugs that haven't surfaced so far and the like. But when everything works out, the feeling of satisfaction is second to none.
There are many interesting demonstrations planned by the other participants of the demonstration team and we look forward to see what others did. But obviously, team AnnieWAY will also be there and we are hard working preparing the demonstrations we want to show. We will show a hand-held stereo reconstruction demonstrator, some videos of the GCDC and obviously some autonomous driving: parking using the Velodyne scanner as sensor and also something new (for us) that came from our work from the GCDC: Following a vehicle autonomously, not only in one dimension, but fully, including steering.
Please make sure to visit us and say hello if you are attending the IV!
By now, we are back at Karlsruhe and have recovered a bit from last week which was exhausting for all of us. The dust is settling, and we are wrapping up the GCDC: the software needs to be brought into a state that can be easily reactivated, some documentation needs to be written and obviously we have still to celebrate our first place with the other scientists at our home institute.
Last week was great fun! We'd like to send greetings and thanks to all the people that we were working with: All the members of the participating teams and all TNO stuff that we met. It was great getting to know you all and we definitively stay in touch. Hopefully, we'll meet before GCDC 2 again!
With the end of the GCDC, this homepage will become a little less active again. We will continue to blog about important developments for Team AnnieWAY, but for now, we all have to get back working on our PhDs for a while.
Results of the GCDC 2011:
- 1st place: AnnieWAY
- 2nd place: Halmstad
- 3rd place: Chalmers
- 4th place: Scoop
- 5th place: Autopia
- 6th place: A-Team
- 7th place: Mekar
- 8th place: Futurum
- 9th place: Latvia
Today was the second (and final) day of the competition. The runs of today were much better than yesterday since some problems of the other days have been fixed. Our car worked well like yesterday and we think that we could make a couple of points during the runs. But let's see, the final result will be presented soon...
At the end of the first competition day our impression of our own performance was quite good. Everything in own car worked as expected and we had some nice runs. However, in one of the runs there was a very dangerous situation not far from a serious accident. Luckily we were not involved in this situation.
In the evening was a barbecue held at TNO and some results of the first day were presented. Unfortunately, the intermediate ranking was disappointing for us since a couple of teams were ranked high which did not meet the safety requirements specified in the rules while other teams were ranked lower which had very reliable systems. However, we will continue tomorrow and see how the final results will look like.
The competition finally was started this morning. After our tests yesterday we were optimistic that our vehicle wouldn't perform badly today. However, we heavily rely on the information provided by the other cars and we already knew that some teams have big problems providing valid and reliable data.
The competition started with two test runs. In both we run into trouble since some predecessing cars were not communicating their position reliable and we could not detect them with our radar system since they were waiting in front of a traffic lights so that the radar did not provide their position. Beside this issue (which we have already raised a couple of times) our car worked properly.
Then, the competition runs started. Luckily, our car did not cause problems and we could drive all heats successfully. However, a dangerous situation occured when another vehicle approached us from behind with high speed and did not break early enough. A collision could only be prevented by manual interaction.
The last day before the event made us a little bit nervous. You see all the things that do not work as you planned and you try to fix as much as possible. However, we did not implement new features today but focussed on testing, testing, and testing. The day started with a test session at the DAF testing site between 6 a.m. and 8 a.m. and continued with testing the following controller and the radar sensor on the A270 highway between Helmond and Eindhoven. Now, our guys are testing the communication with other team vehicles and the platooning controller in parallel. For 7:30 p.m. we ordered pizza since we did not have time for breakfast or lunch today.
Meanwhile, we also obtained the schedule for tomorrow. It is very tight and we will drive 12 heats between 9 a.m. and 5 p.m. We expect it to become a hard day again.
Well, days can become quite long, when you start early. Today was such a day because TNO arranged the possibility to perform platooning tests on the testing site of DAF near Eindhoven. The snag to it was that the test site was free only between 6 a.m. and 8 a.m. so that our four guys who went there had to leave at 5 a.m. already! However, the tests were very fruitful and we could perform successful experiments together with four other teams.
Yesterday evening was the social event of the GCDC. We went to the DAF museum in Eindhoven and watched the old and modern cars and trucks that were shown there. The museum is quite nice and the guides were very motivated in showing the young guys the advantages of the technique of former days.
Meanwhile, we are in Helmond for five days and it is hard to remember which day of the week it currently is. The usual living schedule is lost and the days are formed by working on the software and testing on the car. The only contact to normal life is breakfast in a bakery in the morning (they really offer tasty pastry!) and the dinner in the evening. However, it is difficult to find a restaurant in Helmond that is not yet closed at 9:30 p.m. or 10 p.m. when we usually come back to the hotel.
Fortunately, most of the teams have improved during recent days and we are optimistic that at least seven or eight cars will be working on weekend. Our own software is also improving; we hope to be successful in integrating the radar data by tomorrow. As ususal in software development, debugging takes much longer than you expect...
The focus of our work today was to debug and test the communication with traffic lights and to adapt the behavior of the vehicle so that it reacts appropriately. However, since our vehicle does not see the traffic lights but only reacts on communicated signals we could imitate traffic lights with the communication device of the Audi Q7 while the Passat stopped and accelerated according to the signals received.
Since today, our team is complete since Andreas arrived this afternoon while Holger already arrived yesterday. Hence, all our experts are here and can work on their tasks. For tomorrow, the integration of the radar data will be one of our main jobs.
While we performed many test drives yesterday, today was a silent day in the sense that we only performed some test drives to get data from our radar and to test the platooning controller together with team Futurum. However, our team was busy fixing the bugs in our software that became obvious yesterday and adding the missing functions.
As the final experiment of today we built a platoon of a couple of vehicles and drove autonomously on the A270 highway for approximately 10 kilometers. It was the first test of platooning with several different vehicles at all. As expected some things worked as they should while other things did not yet work, e. g. we lost connection to the lead vehicle twice and had to interact manually to avoid collisions. Hence, enough problems remain to be solved during the next days.
This morning the official vehicle checks were made on a part of the highway, at least for those teams with operating vehicles. The checks included testing the communication, testing the acceleration and deceleration in automatic mode, and testing the switching to manual mode in case of emergency.
We were the first team that had to perform the tests. However, when we arrived at the testing site this morning the communication box of our Passat was not working which caused a lot of stress for Julius and Frank. Finally, we could find a solution of the problem and execute the checks without problem.
|The braking lights are working properly. But who is pushing the braking pedal?|
After arriving yesterday we started our preparation for the challenge on the TNO campus today. A visual inspection of our vehicle had already been done yesterday. Hence, we could focus on testing our following controller again, spending a lot of time in bugfixing the traffic lights module, and creating a map of the highway between Helmond and Eindhoven on which the competition will take place.
Meanwhile, all other teams arrived with their vehicles. The event itself is organized well by TNO offering beverages, snacks, and lunch for free. Lunch was even a Dutch speciality: French fries and meat balls.
Today, we traveled to the competition site at TNO campus in Helmond, the Netherlands. The trip was fine. Since we were four persons, two served as drivers while the other two were programing. The vehicles were communicating with each other all the time during the trip exchanging their position so that we could test the communication over several hours. Moreover, somewhere between Koblenz and Cologne Frank put a second antenna on top of the car (while driving) imitating a road site unit.
At the competition site we were welcome by the organizers and prepared our working place in the huge vehicle testing hall. Meanwhile, four of the other teams (Futurum, Autopia, Mekar, Scoop) arrived. We hope to see the missing teams tomorrow morning.
We are quite busy making the final preparations and adjustments on our software for the GCDC. But still, we managed to squeeze some time out to work on our new Team Video. We finished it today and without further ado, here it is:
The video was not only made with the GCDC in mind, because shortly after it, there is the 2011 IEEE Intelligent Vehicle Symposium, one of the most important conferences on autonomous driving and driver assistance systems. As a special treat, we are also helping to host the conference this year. So we will work extra hard to show our very best there. If you are around, make sure to catch us on June 9th on the Vehicle Demonstration ground where we will show current research demos of our work.
Step by step, we are approaching our goal to complete our software and to add all functions that are needed for the GCDC. Today, we could finish our map matcher software, a tool that can be used to record vehicle tracks, to visualize them and to reuse the tracks as part of a map. Moreover, we implemented the platooning strategy that enables our car to adapt its behavior to a whole convoy, not only on a single vehicle ahead. Test drives were made to check safe operation of the emergency system that enables a safety driver to take over the control of the vehicle in case of danger.
During this week we organized a second training camp for intensive preparation. Already on the first day, we could run an autonomous test drive where our experimental VW Passat followed the Audi Q7 on a test site as well as in normal urban traffic. The Audi Q7 was driven manually while the VW Passat controlled its velocity autonomously. Only steering had to be done by the human driver. Both vehicles were communicating their position, velocity, and acceleration so that the second vehicle could control its behavior based on the position of the first one. But not only for the vehicle was it a new experience but also for us researchers sitting in the second vehicle moving - more or less - autonomously.
Today, we performed another set of experiments testing the performance of our communication devices. Thereto, we placed our two experimental cars at different distances and measured the reliability of communication. From these experiments we can conclude that our communication devices are able to communicate up to 250 meters if the line of sight between the vehicles is not blocked by an obstacle.
The preparations for the GCDC are already started since a couple of month. However, to coordinate work, to fuse the different software modules, and to be able to test the system as a whole we decided to have an intensive working week. In contrast to normal work we moved our workplaces into a single office to be able to exchange ideas, to help each other in debugging, to discuss interfaces, and to assist in driving tests with our two experimental cars.
Yesterday evening, we had a first test with both cars driving through Karlsruhe manually and exchanging position and velocity information continuously by wireless communication. Certainly, not all things did work as they should but we have some ideas how to improve. Unfortunately, some hardware has broken and we have to wait a couple of days until we have the system running again.